Reduction in Late Mortality among 5-Year Survivors of Childhood Cancer

Transcription

1 Original Article Reduction in Late Mortality among 5-Year Survivors of Childhood Cancer Gregory T. Armstrong, M.D., M.S.C.E., Yan Chen, M.M., Yutaka Yasui, Ph.D., Wendy Leisenring, Sc.D., Todd M. Gibson, Ph.D., Ann C. Mertens, Ph.D., Marilyn Stovall, Ph.D., Kevin C. Oeffinger, M.D., Smita Bhatia, M.D., M.P.H., Kevin R. Krull, Ph.D., Paul C. Nathan, M.D., Joseph P. Neglia, M.D., M.P.H., Daniel M. Green, M.D., Melissa M. Hudson, M.D., and Leslie L. Robison, Ph.D. ABSTRACT BACKGROUND Among patients in whom childhood cancer was diagnosed in the and, 18% of those who survived for 5 years died within the subsequent 25 years. In recent decades, cancer treatments have been modified with the goal of reducing life-threatening late effects. METHODS We evaluated late mortality among 34,033 patients in the Childhood Cancer Survivor Study cohort who survived at least 5 years after childhood cancer (i.e., cancer diagnosed before the age of 21 years) for which treatment was initiated during the period from 1970 through The median follow-up was 21 years (range, 5 to 38). We evaluated demographic and disease factors that were associated with death from health-related causes (i.e., conditions that exclude recurrence or progression of the original cancer and external causes but include the late effects of cancer therapy) using cumulative incidence and piecewise exponential models to estimate relative rates and 95% confidence intervals. RESULTS Of the 3958 deaths that occurred during the study period, 1618 (41%) were attributable to health-related causes, including 746 deaths from subsequent neoplasms, 241 from cardiac causes, 137 from pulmonary causes, and 494 from other causes. A reduction in 15-year mortality was observed for death from any cause (from 12.4% in the early to 6.0% in the, P<01 for trend) and from health-related causes (from 3.5% to 2.1%, P<01 for trend). These reductions were attributable to decreases in the rates of death from subsequent neoplasm (P<01), cardiac causes (P<01), and pulmonary causes (P = 4). Changes in therapy according to decade included reduced rates of cranial radiotherapy for acute lymphoblastic leukemia (85% in the, 51% in the, and 19% in the ), of abdominal radiotherapy for Wilms tumor (78%, 53%, and 43%, respectively), of chest radiotherapy for Hodgkin s lymphoma (87%, 79%, and 61%, respectively), and of anthracycline exposure. Reduction in treatment exposure was associated with reduced late mortality among survivors of acute lymphoblastic leukemia and Wilms tumor. From the Departments of Epidemiology and Cancer Control (G.T.A., Y.Y., T.M.G., K.R.K., D.M.G., M.M.H., L.L.R.) and Oncology (G.T.A., M.M.H.), St. Jude Children s Research Hospital, Memphis, TN; the School of Public Health, University of Alberta, Edmonton (Y.C., Y.Y.), and the Hospital for Sick Children, Toronto (P.C.N.) both in Canada; the Cancer Prevention and Clinical Statistics Programs, Fred Hutchinson Cancer Research Center, Seattle (W.L.); the Department of Pediatrics, Emory University, Atlanta (A.C.M.); the Department of Radiation Physics, the University of Texas M.D. Anderson Cancer Center, Houston (M.S.); the Departments of Pediatrics and Medicine, Memorial Sloan Kettering Cancer Center, New York (K.C.O.); the Department of Pediatrics, University of Alabama School of Medicine, Birmingham (S.B.); and the Department of Pediatrics, University of Minnesota Medical School, Minneapolis (J.P.N.). Address reprint requests to Dr. Armstrong at the Department of Epidemiology and Cancer Control, St. Jude Children s Research Hospital, 262 Danny Thomas Pl., Mail Stop 735, Memphis, TN 38105, or at greg. armstrong@ stjude. org. This article was published on January 13, 2016, at NEJM.org. DOI: /NEJMoa Copyright 2016 Massachusetts Medical Society. CONCLUSIONS The strategy of lowering therapeutic exposure has contributed to an observed decline in late mortality among 5-year survivors of childhood cancer. (Funded by the National Cancer Institute and the American Lebanese Syrian Associated Charities.) n engl j med nejm.org 1

2 In the 1960s, fewer than half the children in whom cancer was diagnosed were still alive 5 years later. 1 Now, more than 83% of patients with a childhood cancer in the United States become 5-year survivors of the disease. 2 As a result, in 2013 it was estimated that there were more than 420,000 survivors of childhood cancer in the United States and that by the year 2020 this number would surpass 500, Increased success in the treatment of childhood cancers has been achieved through the systematic conduct of clinical trials to assess the efficacy of multimodal approaches involving combination chemotherapy, radiotherapy, and surgery, along with increased expertise in supportive care. 4,5 Five-year overall survival has been the primary benchmark of therapeutic success. However, as 5-year survival rates increased, it became clear that long-term survivors of childhood cancer were at increased risk for severe and life-threatening therapy-related late effects 6-8 and excess late mortality (i.e., death 5 years after diagnosis) Previous results from the Childhood Cancer Survivor Study (CCSS) showed that by 30 years after the diagnosis of childhood cancer, 18% of 5-year survivors had died. 17 In more recent decades, risk stratification of therapy has increasingly guided the design of treatment regimens for the majority of pediatric cancers. Recent expansion of the CCSS cohort, which now includes survivors in whom cancer was diagnosed from 1970 through 1999, provides an excellent opportunity to evaluate changes over time in therapy and the effect of these changes on overall and cause-specific late mortality. Methods Population The CCSS is a multi-institutional, retrospective, hospital-based cohort study, with longitudinal follow-up of survivors of childhood cancer that was diagnosed and treated at 31 institutions in the United States and Canada ( /ccss.stjude. org). The 34,033 eligible patients in our study included those who had received a diagnosis of cancer before the age of 21 years, with initial treatment performed between January 1, 1970, and December 31, 1999, and who were alive 5 years after the diagnosis. The survivors who were included in the evaluation and who had received the diagnosis of leukemia, tumor of the central nervous system, Hodgkin s lymphoma, non-hodgkin s lymphoma, Wilms tumor, neuroblastoma, soft-tissue sarcoma, and bone tumor represented approximately 20% of childhoodcancer survivors in the United States during the study period. The study methods and design have been described in detail previously. 18 The CCSS was approved by the institutional review board at each of the 31 participating centers. All the participants provided informed consent either in writing or by completing a survey. For participants under the age of 18, consent was provided by a parent or guardian. Ascertainment of Cause of Death The patients who were eligible to participate were included in a search for matching death records in the National Death Index through The index provided underlying and multiple causes of death for deceased patients according to the criteria of the International Classification of Diseases, 9th and 10th Revisions (ICD-9 and ICD-10). For the 139 deaths that predated the National Death Index (i.e., patients who died during the period of 1975 through 1978), we requested death certificates from states in which the deaths had occurred. We used ICD-9 and ICD-10 coding to group the deaths into three mutually exclusive categories: recurrence or progression of the primary cancer; external causes, including accidents, suicides, and poisonings (ICD-9 codes and ICD-10 codes V00 V99, W00 W99, X00 X99, and Y00 Y89); and health-related causes, including subsequent neoplasms (ICD-9 codes and ICD-10 codes C00 C97 and D10 D36), cardiac causes (ICD-9 codes , 402, 404, and and ICD-10 codes I00 I02, I05 I09, I11, I13, I14, I20 I28, and I30 I52), pulmonary causes (ICD-9 codes and ICD-10 codes J00 J99), and all other causes. Data on Cancer Treatments Using standardized CCSS protocols, we abstracted data regarding cancer diagnosis and treatment, including chemotherapy and radiotherapy exposures, from the medical records of 24,243 survivors who provided authorization. 19,20 For the 9790 survivors for whom treatment information was not available, we used multiple-imputation methods. 2 n engl j med nejm.org

3 Late Mortality among Survivors of Childhood Cancer Statistical Analysis Late mortality was evaluated from 5 years after diagnosis until either death or December 31, 2007, the last date of the National Death Index search. We estimated the cumulative incidence of cause-specific death, stratified according to treatment era as defined by 5-year or 10-year intervals and according to the primary cancer diagnosis. We calculated standardized mortality ratios to quantify the rate of death in the CCSS cohort as compared with rates in the U.S. population, according to age, calendar year, and sex. 21 We used multivariable piecewise exponential models to assess relative rates, with 95% confidence intervals, of death from health-related causes in specific treatment eras, as compared with a reference treatment era from 1970 through 1979, after adjustment for sex, primary cancer diagnosis, age at diagnosis, and attained age (which was defined as single-year age segments of the piecewise exponential regression, modeled according to natural cubic splines with knots at 10, 20, 30, and 40 years). We hypothesized that if changes in treatment were responsible for changes in mortality, adjustment for treatment should attenuate the observed effects of treatment eras. Thus, within specific primary cancer groups, we evaluated changes in mortality by comparing the treatment-era effects with and without adjustment for the treatment variables in the model, after adjustment for sex, age at diagnosis, and attained age. To augment the regression-based analysis with a visual description of changes in treatment according to era, we calculated treatment scores for individual survivors from the multivariable piecewise exponential model that included treatment variables, with adjustment for sex, age at diagnosis, and attained age but not with adjustment for treatment era. (Details regarding the derivation and use of the treatment score are provided in Section 1 in the Supplementary Appendix, available with the full text of this article at NEJM.org.) For each survivor with missing treatment data, we used multiple-imputation methods. 22 (Details regarding the methods that were used and an associated sensitivity analysis are provided in Section 2 in the Supplementary Appendix.) For the 440 Canadian patients, for whom the specific cause of death was unknown, we used the predictive-mean-matching method to impute the cause of death, using the age at diagnosis, year of diagnosis, treatment institution, cancer-diagnosis group, and treatment variables as predictors of causes of death. For 27 patients with missing data regarding the time of death, we used the multiple-imputation method proposed by Taylor et al. 23 Because the specific cause of death was not available for Canadian patients, we performed a sensitivity analysis that was restricted to only U.S. survivors. Since no appreciable differences were found, we present the original analysis results. Because of the many comparisons made with these data, P values for differences in mortality among patients with specific cancer types or from specific causes should be regarded as exploratory. Results Study Patients The cohort of 34,033 eligible survivors included more than 9000 survivors who received their initial diagnosis in the, more than 13,000 survivors from the, and more than 11,000 survivors from the (Table 1, and Table S1 in the Supplementary Appendix). These patients provided a total of 705,806 person-years of observation. Of the survivors, 30% were between the ages of 30 and 39 years and 15% were older than 40 years of age at the last follow-up (median age, 28.5 years; range, 5.5 to 58.5). Overall, 57% of the cohort received radiotherapy, including 77% of survivors from the, as compared with only 41% from the. In contrast, more survivors received chemotherapy (including anthracyclines and alkylating agents) in the than in the, but the average cumulative dose of chemotherapeutic agents was lower (Table S2 and Fig. S1 and S2 in the Supplementary Appendix). Rates of Death A total of 3958 deaths occurred in the cohort, with 2002 attributable to recurrence or progression of the primary cancer, 338 to external causes, and 1618 to health-related causes, including 746 from subsequent neoplasms, 241 from cardiac causes, 137 from pulmonary causes, and 494 from other causes. At 15 years after diagnosis, the cumulative incidence of death from any n engl j med nejm.org 3

4 Table 1. Characteristics of 5-Year Survivors of Childhood Cancer. Characteristic All Patients Living Dead no. no. (%) All patients 34,033 30,075 (88.4) 3958 (11.6) Sex Male 18,983 16,628 (87.6) 2355 (12.4) Female 15,050 13,447 (89.3) 1603 (10.7) Race or ethnic group* Non-Hispanic white 21,781 19,575 (89.9) 2206 (10.1) Non-Hispanic black 2,022 1,817 (89.9) 205 (10.1) Hispanic 2,287 2,094 (91.6) 193 (8.4) Other 2,057 1,849 (89.9) 208 (10.1) Unknown 5,886 4,740 (8) 1146 (19.5) Treatment era ,416 7,548 (80.2) 1868 (19.8) ,181 11,699 (88.8) 1482 (11.2) ,436 10,828 (94.7) 608 (5.3) Age at diagnosis (yr) ,463 12,319 (9) 1144 (8.5) 5 9 7,826 6,950 (88.8) 876 (11.2) ,144 6,185 (86.6) 959 (13.4) ,600 4,621 (8) 979 (17.5) Survival after diagnosis (yr) 5 9 4,210 2,349 (55.8) 1861 (44.2) ,298 5,523 (87.7) 775 (12.3) ,285 4,758 (9) 527 (1) ,721 6,343 (94.4) 378 (5.6) ,964 5,692 (95.4) 272 (4.6) ,051 3,924 (96.9) 127 (3.1) 35 1,504 1,486 (98.8) 18 (1.2) Diagnosis Leukemia 10,199 9,019 (88.4) 1180 (11.6) Acute lymphoblastic leukemia 8,500 7,557 (88.9) 943 (11.1) Acute myeloid leukemia 1,222 1,101 (90.1) 121 (9.9) Other leukemia (75.7) 116 (24.3) Hodgkin s lymphoma 4,332 3,647 (84.2) 685 (15.8) Non-Hodgkin s lymphoma 2,837 2,621 (92.4) 216 (7.6) Central nervous system tumor 6,369 5,443 (85.5) 926 (14.5) Astrocytoma 3,904 3,383(86.7) 521 (13.3) Medulloblastoma or PNET 1,380 1,133 (82.1) 247 (17.9) Other 1, (85.4) 158 (14.6) Wilms tumor 3,055 2,898 (94.9) 157 (5.1) Neuroblastoma 2,632 2,457 (93.4) 175 (6.6) Rhabdomyosarcoma 1,679 1,510 (89.9) 169 (10.1) Bone tumor 2,930 2,480 (84.6) 450 (15.4) Ewing s sarcoma (8) 184 (18.5) Osteosarcoma 1,771 1,518 (85.7) 253 (14.3) Other (9) 13 (8.0) * Race or ethnic group was self-reported. PNET denotes primitive neuroectodermal tumor. 4 n engl j med nejm.org

5 Late Mortality among Survivors of Childhood Cancer cause among survivors was 10.7% among those whose disease was diagnosed in the, 7.9% among those whose disease was diagnosed in the, and 5.8% among those whose disease was diagnosed in the (P<01 for the comparison among the three decades) (Fig. 1, and Table S3 and Fig. S1 in the Supplementary Appendix). Across these decades, the cumulative incidence of death from recurrence or progression of a primary cancer decreased, from 7.1% in the to 4.9% in the and 3.4% in the (P<01). Death from health-related causes, which included deaths from late effects of cancer therapy, decreased from 3.1% to 2.4% and 1.9%, respectively (P<01) and reflected significant reductions across the three decades in the rate of death from subsequent malignant neoplasms and from cardiac- and pulmonaryrelated events (Table 2, and Table S4 in the Supplementary Appendix). In a multivariable model adjusted for cancer diagnosis, age at diagnosis, sex, and attained age, more recent treatment eras were associated with a reduced rate of death. The adjusted relative rate per every 5 years differed significantly from for any health-related cause (relative rate, 0.86; 95% confidence interval [CI], 0.82 to 0.89) and for cause-specific death related to subsequent malignant neoplasms (relative rate, 0.83; 95% CI, 0.78 to 0.88), cardiac causes (relative rate, 0.77; 95% CI, 0.68 to 0.86), and pulmonary causes (relative rate, 0.77; 95% CI, 0.66 to 0.89) (Table S5 in the Supplementary Appendix). Similar patterns were seen with respect to reductions in standardized mortality ratios (Table S6 in the Supplementary Appendix). We observed significant reductions across treatment eras in the rate of death from any health-related cause among survivors of acute lymphoblastic leukemia (from 3.2% in the early to 2.1% in the, P<01), Hodgkin s lymphoma (from 5.3% to 2.6%, P = 06), Wilms tumor (from 2.6% to 0.4%, P = 05), and astrocytoma (from 4.7% to 1.8%, P = 2) (Table 2) but not among survivors of the other primary cancer groups. Cardiac mortality declined among patients with acute lymphoblastic leukemia (from 0.6% to 0.1%, P=03), Hodgkin s lymphoma (from 0.9% to %, P = 6), Wilms tumor (from 0.3% to 0%, P = 4), and astrocytoma (from 0.9% to 0%, P = 2), and the rate of death from subsequent neoplasms was significantly reduced among survivors of Wilms tumor (from 1.9% to 0%, P<01). Effects of Changes in Treatment Temporal reductions in exposure to radiotherapy and anthracyclines were observed among patients with acute lymphoblastic leukemia, Hodgkin s lymphoma, Wilms tumor, and astrocytoma (Table S2 and Fig. S2 and S3 in the Supplementary Appendix). Changes in therapy according to decade included reduced rates of cranial radiotherapy for acute lymphoblastic leukemia (85% in the, 51% in the, and 19% in the ), of abdominal radiotherapy for Wilms tumor (78%, 53%, and 43%, respectively), and of chest radiotherapy for Hodgkin s lymphoma (87%, 79%, and 61%, respectively). For the diagnoses of acute lymphoblastic leukemia, Hodgkin s lymphoma, Wilms tumor, and astrocytoma, temporal reductions in 15-year rates of death from health-related causes followed temporal reductions in therapeutic exposure (Fig. 2). The effect of the treatment era on the rate of death from a health-related cause was assessed in multivariable models with and without adjustment for therapy (Table 3, and Table S7 in the Supplementary Appendix). The effect of the treatment era on the relative rate of death from health-related causes was attenuated in models adjusted for therapy for acute lymphoblastic leukemia (unadjusted relative rate, 0.88; adjusted relative rate, 2) and Wilms tumor (relative rate, 0.68 and 0.80, respectively) but not in models adjusted for therapy for Hodgkin s lymphoma (relative rate, 0.79 in the two models) and astrocytoma (relative rate, 0.81 and 0.82, respectively). Discussion Approaches to the treatment of pediatric cancers have evolved during the past five decades with the global objective of achieving increasing rates of cure while minimizing the risk of short-term and long-term toxic effects. 4,5 In this study, we confirmed previously published data showing that patients who were treated in the had a significantly lower rate of death from recurrence or progression of their primary cancer than did patients who were treated in earlier decades. 10,24-26 We now also document the reduced rate of death from treatment-related late effects, such as subsequent cancers and cardio- n engl j med nejm.org 5

6 A Death from Any Cause Yr cumulative mortality, 10.7% ( ) 20, 7.9% ( ), 5.8% ( ) P<01 15 Cumulative Mortality (%) 10 5 Figure 1. All-Cause and Cause-Specific Cumulative Mortality among 5-Year Survivors of Childhood Cancer, According to Decade. Shown is the cumulative incidence of death from any cause (Panel A), from disease recurrence or progression (Panel B), and from any health-related cause (Panel C) among 34,033 patients who survived at least 5 years after childhood cancer for which treatment was initiated during the period from 1970 through The values in parentheses are 95% confidence intervals. The vertical dashed lines indicate 15-year mortality. P values are for the comparisons among the three decades No. at Risk 9,416 13,181 11,436 8,722 13,443 11,411 B Death from Recurrence or Progression Yr cumulative mortality, 7.1% ( ) 20, 4.9% ( ), 3.4% ( ) P<01 15 Cumulative Mortality (%) No. at Risk 10 Years since Diagnosis 8,406 13,105 3,924 8,182 10, Years since Diagnosis 9,416 13,181 11,436 8,722 13,443 11,411 8,406 13,105 3,924 C Death from Health-Related Cause Yr cumulative mortality, 3.1% ( ), 2.4% ( ) 20, 1.9% ( ) P<01 15 Cumulative Mortality (%) No. at Risk 10 8,182 10, Years since Diagnosis 9,416 13,181 11,436 8,722 13,443 11,411 8,406 13,105 3,924 8,182 10, pulmonary conditions. In addition, the results that were generated from the CCSS cohort provide evidence that the strategy of reducing treatment exposure in order to decrease the frequency of late effects is translating into a significant reduction in observed late mortality and an extension of the life span of children and adolescents who are successfully treated for cancer. Appreciation of the risk of long-term adverse consequences of therapy 6-8 resulted in the design and testing of newer treatment regimens to reduce the potential for late effects. This was generally achieved by a reduction in therapeutic exposures in patients who were considered to be at low risk for recurrence of the primary cancer, while providing therapy that would maintain or improve long-term disease-free survival. 14,27-35 The availability of treatment-exposure data in the CCSS cohort, including cumulative doses of most chemotherapeutic regimens and organ-specific radiotherapy dosimetry, provided an excellent opportunity to evaluate whether the risk reduction that had been observed in more recent treatment eras was directly associated with a reduction in therapeutic exposure. We observed temporal reductions in health-related mortality concurrent with a reduction in therapeutic exposure among patients with acute lymphoblastic leukemia and Wilms tumor. However, with respect to Hodgkin s lymphoma and astrocytoma, our findings suggest that factors other than reduced treatment exposures may have caused the observed reductions in late health-related mortality. Potential contributors to decreased late mortality include increased use and accuracy of screening methods Although data suggest that guideline-based screening and care have not been universally adopted, 39 it should be expected that these efforts would have a positive effect on 6 n engl j med nejm.org

7 Late Mortality among Survivors of Childhood Cancer Table 2. Cumulative Incidence of Death at 15 Years after Primary Cancer Diagnosis among 5-Year Survivors.* Diagnosis and Year Death from Any Cause Death from Recurrence or Progression Any Subsequent Neoplasm Death from Health-Related Cause percentage Cardiac Cause Pulmonary Cause Other All diagnoses Year of diagnosis P value <01 <01 <01 < Acute lymphoblastic leukemia Year of diagnosis P value <01 <01 < <01 Hodgkin s lymphoma Year of diagnosis P value <01 < Wilms tumor Year of diagnosis P value < Astrocytoma Year of diagnosis P value < * All P values are for testing for a trend from to Other health-related causes are those not included in the listed categories. The total numbers of survivors were 3696 in , 5720 in , 7184 in , 5997 in , and 5442 in n engl j med nejm.org 7

8 A Acute Lymphoblastic Leukemia B Hodgkin s Lymphoma Treatment Score Yr Cumulative Incidence (%) Treatment Score Yr Cumulative Incidence (%) Treatment Era Treatment Era C Wilms Tumor Treatment Score Yr Cumulative Incidence (%) D Astrocytoma Treatment Score Yr Cumulative Incidence (%) Treatment Era Treatment Era Figure 2. Treatment Score and 15-Year Cumulative Mortality, According to Type of Cancer and Decade. Shown are box plots of treatment scores (left y axis) and 15-year cumulative mortality (red dot, right y axis) among 5-year survivors of childhood acute lymphoblastic leukemia (Panel A), Hodgkin s lymphoma (Panel B), Wilms tumor (Panel C), and astrocytoma (Panel D), according to the decade during which treatment was received. Treatment scores quantify the treatment-associated propensity for late mortality from health-related causes. Scores are standardized so that those from the have a mean of 0 and a standard deviation of. The reductions in treatment scores that are shown over the decades indicate a reduced treatment-associated propensity for late death. The horizontal line in each box plot shows the median, and the bottom and top of the box are located at the 25th and 75th percentiles, respectively. The I bars represent values that are more than times the interquartile range from the border of each box. health-related mortality. In addition, during the past several decades, there have been improvements in medical care that may delay or prevent death from the late effects of therapy. However, our ability to directly measure these changes in the CCSS population is limited. It should also be 8 n engl j med nejm.org

9 Late Mortality among Survivors of Childhood Cancer Table 3. Relative Rates of Death from Health-Related Causes and the Effect of Treatment Exposure among 5-Year Survivors of Specific Childhood Cancers per 5-Year Treatment Era, According to Multivariable Model.* Treatment Era Acute Lymphoblastic Leukemia Hodgkin s Lymphoma Wilms Tumor Astrocytoma relative rate (95% CI) No adjustment for therapy 0.88 ( ) 0.79 ( ) 0.68 (6 0.84) 0.81 ( ) Adjustment for therapy 2 ( ) 0.79 ( ) 0.80 (9 8) 0.82 ( ) * All models were adjusted for sex, age at diagnosis, and attained age. CI denotes confidence interval. Data were adjusted for cranial radiotherapy dose, anthracycline dose, and exposure to epipodophyllotoxins and glucocorticoids. Data were adjusted for chest-directed radiotherapy dose, anthracycline dose, cyclophosphamide-equivalent dose, and splenectomy. Data were adjusted for abdominal radiotherapy dose and anthracycline dose. Data were adjusted for cranial radiotherapy dose and exposure to chemotherapy. noted that for certain cancers, primarily neuroblastoma, there has been an increase in late mortality in more recent decades, presumably attributable to increased therapeutic intensity that has resulted in improved 5-year survival but an increased risk of late effects and delayed recurrence or progression of the primary cancer. The overall decrease in all-cause mortality is attributable primarily to a reduction in the rate of death from recurrence or progression of primary cancers, a finding that is consistent with that in previous studies, 12,25 and suggests that 5-year survivors in more recent eras have more durable remissions or a more favorable response to therapy for relapse or recurrence of their primary cancers. Combined improvements in the treatment of primary cancers and reductions in health-related mortality have resulted in a relative reduction of almost 50% in all-cause late mortality among survivors of childhood cancer (from 10.7% at 15 years after diagnosis among survivors from the to 5.8% among survivors from the ). Our previous study of registry data suggested that survivors in recent eras may be at a lower risk for death from late effects of cancer therapy than were survivors in earlier eras. 24 The large size of the CCSS cohort and its detailed treatment information provided compelling evidence of a reduction in rates of death from subsequent neoplasm, cardiac causes, and pulmonary causes over the course of three decades. However, any evaluation of our current findings needs to consider several limitations of our study. First, the outcome of health-related causes of death does not allow direct attribution of deaths to sequelae from the treatment of childhood cancer. Second, we could not quantify and directly consider temporal changes in medical care as part of our review. Third, there was a potential for bias resulting from shorter follow-up of survivors from the (although 10-year patterns appeared to be consistent with those for 15-year mortality [Table S8 in the Supplementary Appendix]). Fourth, we did not evaluate temporal trends in the incidence of specific treatmentrelated chronic health conditions that could increase the risk of death. In conclusion, in our study of long-term survivors of pediatric cancer, we confirmed the effect of treatment regimens that have been designed to reduce the potential risk and severity of late effects. Quantitative evidence now shows that the modification of treatment regimens to reduce radiotherapy and chemotherapy exposures, along with increased promotion of strategies for early detection of late effects and improvements in medical care for late effects of therapy, has resulted in the extension of life spans for many survivors of childhood cancer. Presented at the plenary session of the Annual Meeting of the American Society of Clinical Oncology, Chicago, May 29 June 2, Supported by a grant from the National Cancer Institute (CA55727, to Dr. Armstrong), by a grant to St. Jude Children s Research Hospital from the Cancer Center Support Grants (CA21765), and by the American Lebanese Syrian Associated Charities. Disclosure forms provided by the authors are available with the full text of this article at NEJM.org. n engl j med nejm.org 9

10 References 1. Ries LG, Pollack ES, Young JL Jr. Cancer patient survival: Surveillance, Epidemiology, and End Results Program, J Natl Cancer Inst 1983; 70: SEER Cancer Statistics Review (CSR), Bethesda, MD: National Cancer Institute ( csr/ 1975_2012/ ). 3. Robison LL, Hudson MM. Survivors of childhood and adolescent cancer: lifelong risks and responsibilities. Nat Rev Cancer 2014; 14: Hudson MM, Neglia JP, Woods WG, et al. Lessons from the past: opportunities to improve childhood cancer survivor care through outcomes investigations of historical therapeutic approaches for pediatric hematological malignancies. Pediatr Blood Cancer 2012; 58: Green DM, Kun LE, Matthay KK, et al. Relevance of historical therapeutic approaches to the contemporary treatment of pediatric solid tumors. Pediatr Blood Cancer 2013; 60: Oeffinger KC, Mertens AC, Sklar CA, et al. Chronic health conditions in adult survivors of childhood cancer. N Engl J Med 2006; 355: Armstrong GT, Kawashima T, Leisenring W, et al. Aging and risk of severe, disabling, life-threatening, and fatal events in the Childhood Cancer Survivor Study. J Clin Oncol 2014; 32: Hudson MM, Ness KK, Gurney JG, et al. Clinical ascertainment of health outcomes among adults treated for childhood cancer. JAMA 2013; 309: Mertens AC, Yasui Y, Neglia JP, et al. Late mortality experience in five-year survivors of childhood and adolescent cancer: the Childhood Cancer Survivor Study. J Clin Oncol 2001; 19: Hudson MM, Jones D, Boyett J, Sharp GB, Pui CH. Late mortality of long-term survivors of childhood cancer. J Clin Oncol 1997; 15: Mertens AC, Liu Q, Neglia JP, et al. Cause-specific late mortality among 5-year survivors of childhood cancer: the Childhood Cancer Survivor Study. J Natl Cancer Inst 2008; 100: Möller TR, Garwicz S, Barlow L, et al. Decreasing late mortality among five-year survivors of cancer in childhood and adolescence: a population-based study in the Nordic countries. J Clin Oncol 2001; 19: Li FP, Myers MH, Heise HW, Jaffe N. The course of five-year survivors of cancer in childhood. J Pediatr 1978; 93: Pui CH, Cheng C, Leung W, et al. Extended follow-up of long-term survivors of childhood acute lymphoblastic leukemia. N Engl J Med 2003; 349: Reulen RC, Winter DL, Frobisher C, et al. Long-term cause-specific mortality among survivors of childhood cancer. JAMA 2010; 304: Cardous-Ubbink MC, Heinen RC, Langeveld NE, et al. Long-term causespecific mortality among five-year survivors of childhood cancer. Pediatr Blood Cancer 2004; 42: Armstrong GT, Liu Q, Yasui Y, et al. Late mortality among 5-year survivors of childhood cancer: a summary from the Childhood Cancer Survivor Study. J Clin Oncol 2009; 27: Robison LL, Armstrong GT, Boice JD, et al. The Childhood Cancer Survivor Study: a National Cancer Institute-supported resource for outcome and intervention research. J Clin Oncol 2009; 27: Leisenring WM, Mertens AC, Armstrong GT, et al. Pediatric cancer survivorship research: experience of the Childhood Cancer Survivor Study. J Clin Oncol 2009; 27: Stovall M, Weathers R, Kasper C, et al. Dose reconstruction for therapeutic and diagnostic radiation exposures: use in epidemiological studies. Radiat Res 2006; 166: Department of Health and Human Services, Centers for Disease Control and Prevention, National Center for Health Statistics. Compressed mortality file on CDC Wonder Online Database ( wonder.cdc.gov/ wonder/ help/ cmf.html). 22. Little RJA, Rubin DB. Statistical analysis with missing data. 2nd ed. New York: John Wiley, Taylor JM, Muñoz A, Bass SM, Saah AJ, Chmiel JS, Kingsley LA. Estimating the distribution of times from HIV seroconversion to AIDS using multiple imputation. Stat Med 1990; 9: Armstrong GT, Pan Z, Ness KK, Srivastava D, Robison LL. Temporal trends in cause-specific late mortality among 5-year survivors of childhood cancer. J Clin Oncol 2010; 28: Garwicz S, Anderson H, Olsen JH, et al. Late and very late mortality in 5-year survivors of childhood cancer: changing pattern over four decades experience from the Nordic countries. Int J Cancer 2012; 131: Möller TR, Garwicz S, Perfekt R, et al. Late mortality among five-year survivors of cancer in childhood and adolescence. Acta Oncol 2004; 43: Pui CH, Campana D, Pei D, et al. Treating childhood acute lymphoblastic leukemia without cranial irradiation. N Engl J Med 2009; 360: Sullivan MP, Chen T, Dyment PG, Hvizdala E, Steuber CP. Equivalence of intrathecal chemotherapy and radiotherapy as central nervous system prophylaxis in children with acute lymphatic leukemia: a Pediatric Oncology Group study. Blood 1982; 60: Conklin HM, Krull KR, Reddick WE, Pei D, Cheng C, Pui CH. Cognitive outcomes following contemporary treatment without cranial irradiation for childhood acute lymphoblastic leukemia. J Natl Cancer Inst 2012; 104: Donaldson SS, Link MP. Combined modality treatment with low-dose radiation and MOPP chemotherapy for children with Hodgkin s disease. J Clin Oncol 1987; 5: Schwartz CL, Constine LS, Villaluna D, et al. A risk-adapted, response-based approach using ABVE-PC for children and adolescents with intermediate- and highrisk Hodgkin lymphoma: the results of P9425. Blood 2009; 114: D Angio GJ, Breslow N, Beckwith JB, et al. Treatment of Wilms tumor: results of the Third National Wilms Tumor Study. Cancer 1989; 64: Lipshultz SE, Colan SD, Gelber RD, Perez-Atayde AR, Sallan SE, Sanders SP. Late cardiac effects of doxorubicin therapy for acute lymphoblastic leukemia in childhood. N Engl J Med 1991; 324: Green DM, Breslow NE, Beckwith JB, et al. Effect of duration of treatment on treatment outcome and cost of treatment for Wilms tumor: a report from the National Wilms Tumor Study Group. J Clin Oncol 1998; 16: Packer RJ, Ater J, Allen J, et al. Carboplatin and vincristine chemotherapy for children with newly diagnosed progressive low-grade gliomas. J Neurosurg 1997; 86: Henderson TO, Amsterdam A, Bhatia S, et al. Systematic review: surveillance for breast cancer in women treated with chest radiation for childhood, adolescent, or young adult cancer. Ann Intern Med 2010; 152: ; W Hudson MM, Leisenring W, Stratton KK, et al. Increasing cardiomyopathy screening in at-risk adult survivors of pediatric malignancies: a randomized controlled trial. J Clin Oncol 2014; 32: Kremer LC, Mulder RL, Oeffinger KC, et al. A worldwide collaboration to harmonize guidelines for the long-term followup of childhood and young adult cancer survivors: a report from the International Late Effects of Childhood Cancer Guideline Harmonization Group. Pediatr Blood Cancer 2013; 60: Nathan PC, Ford JS, Henderson TO, et al. Health behaviors, medical care, and interventions to promote healthy living in the Childhood Cancer Survivor Study cohort. J Clin Oncol 2009; 27: Copyright 2016 Massachusetts Medical Society. 10 n engl j med nejm.org